NEON User Guide to Herbaceous Clip Harvest (DP1.10023.001) Date: 05/11/2021 Author: Katherine Jones Revision: F

Title: NEON User Guide to Herbaceous clip harvest (DP1.10023.001) Date: 05/11/2021 Author: Katherine Jones Revision: F

NEON USER GUIDE TO HERBACEOUS CLIP HARVEST (DP1.10023.001)

PREPARED BY ORGANIZATION Katherine Jones SCI Courtney Meier SCI Dave Barne SCI Title: NEON User Guide to Herbaceous clip harvest (DP1.10023.001) Date: 05/11/2021 Author: Katherine Jones Revision: F

CHANGE RECORD

REVISION DATE DESCRIPTION OF CHANGE A 06/16/2017 Inial Release B 11/15/2018 2018 sampling modiﬁcaons described Addional known data quality issue described, BoutNumber described C 10/14/2019 for Distributed plots Included general statement about usage of neonUlies R package and D 06/22/2020 statement about possible locaon changes E 10/19/2020 Added Sampling Design Changes secion and content Adjusted mechanism for generang eventID and enhanced tracking of F 05/10/2021 plots not sampled according to protocol guidelines Title: NEON User Guide to Herbaceous clip harvest (DP1.10023.001) Date: 05/11/2021 Author: Katherine Jones Revision: F

TABLE OF CONTENTS

1 DESCRIPTION 1 1.1 Purpose ...... 1 1.2 Scope ...... 1

2 RELATED DOCUMENTS AND ACRONYMS 2 2.1 Associated Documents ...... 2 2.2 Acronyms ...... 2

3 DATA PRODUCT DESCRIPTION 3 3.1 Spaal Sampling Design ...... 3 3.2 Temporal Sampling Design ...... 6 3.3 Sampling Design Changes ...... 7 3.4 Variables Reported ...... 8 3.5 Spaal Resoluon and Extent ...... 9 3.6 Temporal Resoluon and Extent ...... 10 3.7 Associated Data Streams ...... 10 3.8 Product Instances ...... 10 3.9 Data Relaonships ...... 11

4 DATA QUALITY 11 4.1 Data Entry Constraint and Validaon ...... 11 4.2 Automated Data Processing Steps ...... 12 4.3 Data Revision ...... 12 4.4 Quality Flagging ...... 13

LIST OF TABLES AND FIGURES

Table 1 Descripons of the dataQF codes for quality ﬂagging ...... 13

Figure 1 A 20m x 20m (400m2) NEON plot showing potenal clip harvest ”cells” (dashed blue lines). Cells overlapping 10m2 nested subplots are not sampled for herbaceous biomass. When grazing management is absent, one randomly selected grid cell is chosen for harvest per sampling bout. At sites managed for grazing, two grid cells are sampled per bout. The pointID (red star) is used to calculate the locaon in UTM of a given Clip List coordinate (red ‘x’)...... 5

i Title: NEON User Guide to Herbaceous clip harvest (DP1.10023.001) Date: 05/11/2021 Author: Katherine Jones Revision: F

Figure 2 A 40m x 40m (1600 m2) NEON Tower plot showing clip-harvest cells within a 400 m2 subplot. Cells overlapping 10m2 nested subplots are not sampled for herbaceous biomass. Clip harvest samples are generated from two randomly selected subplots per sampling bout, but for clarity, cells are not shown in 3 of 4 subplots. At sites managed for grazing, two grid cells per selected subplot are sampled per bout. For each subplotID (grey numbers), the pointID at the SW corner (red stars) is used to calculate the locaon in UTM of a given Clip List coordinate (red ‘x’)...... 6 Figure 3 Schemac of the applicaons used by ﬁeld technicians to enter herbaceous biomass data ...... 12

ii Title: NEON User Guide to Herbaceous clip harvest (DP1.10023.001) Date: 05/11/2021 Author: Katherine Jones Revision: F

1 DESCRIPTION

1.1 Purpose

This document provides an overview of the data included in this NEON Level 1 data product, the quality controlled product generated from raw Level 0 data, and associated metadata. In the NEON data products framework, the raw data collected in the ﬁeld, for example, the dry weights of lier funconal groups from a single collecon event are considered the lowest level (Level 0). Raw data that have been quality checked via the steps detailed herein, as well as simple metrics that emerge from the raw data are considered Level 1 data products. The text herein provides a discussion of measurement theory and implementaon, data product prove- nance, quality assurance and control methods used, and approximaons and/or assumpons made during L1 data creaon.

1.2 Scope

This document describes the steps needed to generate the L1 data product ‘Herbaceous clip harvest’ - the dry weight of annual producon of herbaceous plants clipped from randomly placed clip cells in terrestrial plots - and associated metadata from input data. This document also provides details relevant to the publicaon of the data products via the NEON data portal, with addional detail available in the file, NEON Data Publicaon Workbook for TOS Herbaceous Biomass: QA/QC of Raw Field and Lab Data (DP1.10023.001) (AD[05]), provided in the download package for this data product. This document describes the process for ingesng and performing automated quality assurance and control procedures on the data collected in the field pertaining to TOS Protocol and Procedure: Measure- ment of Herbaceous Biomass (AD[07]). The raw data that are processed in this document are detailed in the file, NEON Raw Data Ingest Workbook for TOS Herbaceous Biomass (DP0.10023.001) (AD[04]), provided in the download package for this data product. Please note that raw data products (denoted by ‘DP0’) may not always have the same numbers (e.g., ‘10033’) as the corresponding L1 data product.

Page 1 of 13 Title: NEON User Guide to Herbaceous clip harvest (DP1.10023.001) Date: 05/11/2021 Author: Katherine Jones Revision: F

2 RELATED DOCUMENTS AND ACRONYMS

2.1 Associated Documents

AD[01] NEON.DOC.000001 NEON Observatory Design (NOD) Requirements AD[02] NEON.DOC.000913 TOS Science Design for Spaal Sampling AD[03] NEON.DOC.002652 NEON Data Products Catalog Available with data AD[04] Validaon csv download Available with data AD[05] Variables csv download AD[06] NEON.DOC.000914 TOS Science Design for Plant Biomass and Producvity TOS Protocol and Procedure: Measurement of Herbaceous AD[07] NEON.DOC.014037 Biomass AD[08] NEON.DOC.000913 TOS Science Design for Spaal Sampling AD[09] NEON.DOC.000913 TOS Science Design for Plant Diversity AD[10] NEON.DOC.000008 NEON Acronym List AD[11] NEON.DOC.000243 NEON Glossary of Terms TOS Protocol and Procedure: Measurement of Herbaceous AD[12] NEON.DOC.014037 Biomass NEON Algorithm Theorecal Basis Document: OS Generic Transi- AD[13] NEON.DOC.004825 ons Available on NEON AD[14] NEON Ingest Conversion Language Funcon Library data portal Available on NEON AD[15] NEON Ingest Conversion Language data portal Available with data AD[16] Categorical Codes csv download

2.2 Acronyms

Acronym Deﬁnion ANPP Aboveground Net Primary Producvity TOS Terrestrial Observaon System

Page 2 of 13 Title: NEON User Guide to Herbaceous clip harvest (DP1.10023.001) Date: 05/11/2021 Author: Katherine Jones Revision: F

3 DATA PRODUCT DESCRIPTION

The herbaceous clip harvest data products provide mass data for herbaceous plant funconal groups from individual sampling bouts. All masses reported following processing are reported at the spaal resoluon of one clip strip and the temporal resoluon of a single collecon event (Figure 1). Mass data for each collecon event are reported to an accuracy of 0.01 grams. Masses < 0.01g are also reported and indicate presence of an idenfiable sample present at masses below accuracy of the scales used to weigh samples. For a minimum of one bout per growing season, the aboveground dry mass producon associated with key herbaceous funconal groups will be provided. These funconal groups are defined as: • Bryophytes for which annual producon is discernable • C3 cool-season graminoids • C4 warm-season graminoids • Nitrogen-fixing forbs and woody species at decimeter height < 1 cm • Non-leguminous forbs • Woody-stemmed shrubs with diameter at decimeter height < 1 cm All data products include spaal and temporal sampling details, as well as required metadata. For addional informaon on the sampling design and the associated protocol, see TOS Science Design for Plant Biomass, Producvity, and Leaf Area Index (AD[06]), and the TOS Field and Lab Protocol for Herbaceous Plant Biomass (AD[07]). Following data collecon in the field and the lab, several standard QA/QC procedures are implemented, as described in Secon 5 of this document, in order to produce quality-checked Level 1 herbaceous clip harvest data products. Herbaceous clip harvest data may be used to esmate annual Aboveground Net Primary Producvity (ANPP) and aboveground biomass at plot, site, and connental scales. They also provide essenal data for understanding vegetave carbon fluxes over me.

3.1 Spaal Sampling Design

The Herbaceous Clip Harvest sampling is implemented in Tower plots at all terrestrial NEON sites and in Distributed plots at terrestrial NEON sites with non-forest, herbaceous dominated, Naonal Land Cover Database (NLCD) Classificaon. Sampling occurs at a subset of Tower plots at some sites and at all Tower plots at others, and in a maximum of twenty randomly selected Distributed plots, with ≥ 50% herbaceous plant cover, per site. In Distributed plots, the maximum potenal sampling effort (n=20 plots) corresponds to sampling in ½ to ⅔ of the established Distributed plots per site. The Distributed plots at which the herbaceous clip harvest protocol is performed are chosen according to a spaally balanced, strafied-random design in order to create an unbiased esmate of herbaceous biomass at the site scale (see AD[08] for details). In forest, grassland, and rangeland systems, for both plot types, the clip harvest is performed in 0.1 m x 2 m clip strips that exist within 0.5 m x 3 m gridded clip “cells” (Figure 1 and Figure 2). For typical sampling bouts, one gridded clip-area “cell” is randomly chosen per 400 m2 plot or subplot. However, at sites

Page 3 of 13 Title: NEON User Guide to Herbaceous clip harvest (DP1.10023.001) Date: 05/11/2021 Author: Katherine Jones Revision: F

managed for grazing, an addional grid cell is placed within a grazing exclosure, resulng in two clip-strips harvested per sampling bout. Grid cells that overlap with 25m2 nested subplots (used for seedling and sapling counts) may be used for clip harvests, but those that overlap with 1m2 and 10m2 nested subplots (used for diversity sampling) are never clipped. This ensures that generaon of herbaceous plant producvity data minimally affects the quality of plant diversity data. See TOS Science Design for Plant Biomass and Producvity (AD[06]), TOS Protocol and Procedure: Measurement of Herbaceous Biomass (AD[07]), TOS Science Design for Spaal Sampling (AD[08]) and TOS Science Design for Plant Diversity (AD[09]) for further details. Placement of clip strips is random and ulizes the randomized list of grid cell locaons being ulized for herbaceous clip harvest (AD[12]). This is an ordered list of all available sampling locaons within a given plot, from which a clipCell is removed aer being ulized for sampling in any one of the three sampling protocols. Removal precludes consideraon of a clipCell for future sampling under all protocols. Clip strip dimensions and orientaon at agricultural sites differs from clip strips at non-agricultural sites. In all agricultural systems, clip strips are oriented perpendicular to crop rows. Plots planted with barley and wheat ulize clip sps that are 2.0 m x 0.5 m; plots planted with corn, sorghum and soybeans clip biomass within 1.5 m x 0.65 m clip strips consistent with Long-Term Agroecosystem Research (LTAR) protocols. As much as possible, sampling occurs in the same locaons over the lifeme of the Observatory. However, over me some sampling locaons may become impossible to sample, due to disturbance or other local changes. When this occurs, the locaon and its locaon ID are rered. A locaon may also shi to slightly different coordinates. Refer to the locaons endpoint of the NEON API for details about locaons that have been moved or rered: https://data.neonscience.org/data-api/endpoints/locations/

Page 4 of 13 Title: NEON User Guide to Herbaceous clip harvest (DP1.10023.001) Date: 05/11/2021 Author: Katherine Jones Revision: F

Figure 1: A 20m x 20m (400m2) NEON plot showing potenal clip harvest ”cells” (dashed blue lines). Cells overlapping 10m2 nested subplots are not sampled for herbaceous biomass. When grazing management is absent, one randomly selected grid cell is chosen for harvest per sampling bout. At sites managed for grazing, two grid cells are sampled per bout. The pointID (red star) is used to calculate the locaon in UTM of a given Clip List coordinate (red ‘x’).

Page 5 of 13 Title: NEON User Guide to Herbaceous clip harvest (DP1.10023.001) Date: 05/11/2021 Author: Katherine Jones Revision: F

Figure 2: A 40m x 40m (1600 m2) NEON Tower plot showing clip-harvest cells within a 400 m2 subplot. Cells overlapping 10m2 nested subplots are not sampled for herbaceous biomass. Clip harvest samples are generated from two randomly selected subplots per sampling bout, but for clarity, cells are not shown in 3 of 4 subplots. At sites managed for grazing, two grid cells per selected subplot are sampled per bout. For each subplotID (grey numbers), the pointID at the SW corner (red stars) is used to calculate the locaon in UTM of a given Clip List coordinate (red ‘x’).

3.2 Temporal Sampling Design

NEON field and lab technicians will generate the herbaceous clip harvest data product at least once per growing season for each site. However, the temporal sampling resoluon depends on plot type. Dis- tributed plots are sampled at five year intervals at each site. In a site sampling year, Distributed plots are harvested once per growing season except for agriculture sites, which can be harvested more frequently when mulple crops are planted in a single growing season. The default sampling frequency for Tower plots is once per year, but the following factors may result in addional sampling bouts per growing season, and hence addional instances of the data product, and are assessed on a per site basis: • Sites with disnct early-season (C3 plants) and late-season biomass peaks (C4 plants) are clip- harvested twice per growing season. • Sites acvely managed for grazing are clip-harvested every 4 or 8 weeks depending on the compen- satory response of species present. • Sites with connuous growth and decomposion (e.g., tropical and sub-tropical grasslands) are harvested more frequently than sites with periods of senescence. Within a sampling season, herbaceous clip harvests are scheduled at phenologically consistent mes across sites to allow for data comparability across the Observatory and through me. To determine phenologically consistent sampling dates across widely different ecosystems, the TOS ulizes MODIS-EVI phe-

Page 6 of 13 Title: NEON User Guide to Herbaceous clip harvest (DP1.10023.001) Date: 05/11/2021 Author: Katherine Jones Revision: F

nology data for the majority of sites (Didan 2015), and more specifically, per site averages from the most recent 10 years that are updated every 5 years. Ungrazed, wildland site herbaceous clip harvest sampling schedules correspond to the average date aer which greenness begins to decrease, ensuring a high probability that all herbaceous biomass has been produced prior to clipping. For grazed sites, scheduled sampling corresponds to green-up dates, and it is assumed that cows are present from that point forward. At agriculture sites, MODIS is considered a rough guide; field staff monitor crops for senescence and begin clipping aer the first leaves begin to dry or yellow, and work with sites to ensure clipping occurs before crops are harvested. Clip harvest events and the resulng herbaceous biomass values can be grouped by bout according to a unique eventID. The eventID is comprised of HBP (the data product code), year, site code, and an integer - the Internaonal Organizaon for Standardizaon (ISO) week date in which the sampling bout began.

3.3 Sampling Design Changes

Several design changes have been implemented over the course of data collecon. Such changes arise due to connual evaluaon of the sampling design for best pracces in collaboraon with external techni- cal working groups. Design changes also occur when opmizaon is necessary to ensure that allocaon of sampling effort is poised to maximize returns to the scienfic community. • Prior to 2018: Distributed plots were sampled at three-year intervals. Subsequent sampling at five- year intervals coincides with other plant biomass and producvity data products at each site. • Prior to 2018: All Tower plots were sampled during each sampling bout at each site. • Prior to 2018: The Tower plot intra-annual sampling at grazed sites occurred every four weeks during the growing season. • Prior to 2018: At select grazed sites, those bouts not associated with peak biomass during which samples are not sorted by funconal group, the complete clip strip was collected and weighed. For the 2018 and subsequent sampling years a subsample of the plot-specific clip harvest is dried and weighed, but data-driven esmates of the enre sample mass are reported. • Prior to 2019: Samples were sorted into funconal groups that included the category ‘Leguminous Forbs (LFB)’. Subsequent sampling places leguminous forbs and nitrogen fixers < 1 cm DBH into a ‘N-fixer (NFX)’ funconal group category. • 2020: The field samplingImpraccal was added to the data to allow for the generaon of a record when a plot could not be sampled or a sample could not be dried for a parcular bout and year. If field sampling was not possible samplingImpraccal is populated with a value other than ‘OK’ (e.g., ‘locaon flooded’) in the hbp_perbout table, no sample is collected, and no record is generated in the hbp_massdata table. If a sample is collected but not dried or weighed, samplingImpraccal is populated with a value other than ‘OK’ in the hbp_massdata table and no mass data are available for that sample. • 2020: The field biophysicalCriteria was added to the data to account for instances when sampling occurred but condions were not opmal. If condions were subopmal - the sample was not col-

Page 7 of 13 Title: NEON User Guide to Herbaceous clip harvest (DP1.10023.001) Date: 05/11/2021 Author: Katherine Jones Revision: F

lected during peak biomass - the field biophysicalCriteria is populated with a value other than ‘OK - no known excepons’ (e.g., ‘condions not met: most plants not yet flowering’) but the sample is collected and mass data are generated. • Prior to 2021: The bout number for each clipping event within a site was assigned a number. Sam- pling at Tower plots began with bout = 1 and connued with ordered numeric increments for each sampling event at each site within a sampling year. Bout number = 0 was assigned to Distributed plots that are typically harvested only once per season for biomass stocks. This means that agriculture sites where different Distributed Plots may be harvested at different mes all have boutNum- ber = 0, regardless of clipDate. • Prior to 2021: The eventID was comprised of HBP, year, site code, and the bout number. • May of 2021: The data were structured such that records were created when sampling at possible and scheduled plots did not occur. Such instances include: -Distributed plots in forested NLCD types. The field targetTaxaPresent is set to ‘No’, and the remarks field is populated with ‘forest cover type, sampling criteria not met.’ -Distributed plots in non-forest NLCD types where the herbaceous cover < 50%. The field targetTaxaPre- sent is set to ‘No’, and the remarks field is populated with ‘herbaceous cover < 50% of plot area, sampling criteria not met.’ -Grazed sites when a plot was not subjected to grazing during the interval prior to the scheduled sampling event. The field samplingImpraccal is set to ‘exclosure cell not clipped, livestock absent’ or ‘ambient cell not clipped, livestock absent’ for the clip cell under the exclosure and exposed to grazing respecvely.

3.4 Variables Reported

All variables reported from the field or laboratory technician (L0 data) are listed in the file, NEON Raw Data Ingest Workbook for TOS Herbaceous Biomass (DP0.10023.001) (AD[04]). All variables reported in the published data (L1 data) are also provided separately in the file, NEON Data Publicaon Workbook for TOS Herbaceous Biomass: QA/QC of Raw Field and Lab Data (DP1.10023.001) (AD[05]). Field names have been standardized with Darwin Core terms (http://rs.tdwg.org/dwc/; accessed 16 February 2014), the Global Biodiversity Informaon Facility vocabularies (http://rs.gbif.org/vocabulary/ gbif/; accessed 16 February 2014), the VegCore data diconary (https://projects.nceas.ucsb.edu/nceas/ projects/bien/wiki/VegCore; accessed 16 February 2014), where applicable. NEON TOS spaal data em- ploys the World Geodec System 1984 (WGS84) for its fundamental reference datum and GEOID09 for its reference gravitaonal ellipsoid. Latudes and longitudes are denoted in decimal notaon to six decimal places, with longitudes indicated as negave west of the Greenwich meridian. Some variables described in this document may be for NEON internal use only and will not appear in downloaded data.

Page 8 of 13 Title: NEON User Guide to Herbaceous clip harvest (DP1.10023.001) Date: 05/11/2021 Author: Katherine Jones Revision: F

3.5 Spaal Resoluon and Extent

The overall spaal hierarchy of the herbaceous clip harvest data product is subtly different for Tower plots compared to Distributed plots. For Tower plots that are 40 m x 40 m and larger, the gridded clip harvest locaons exist within subplotIDs within each plotID (Figure 2): Tower plots (≥ 1600m2): clipID ¨ subplotID ¨ plotID ¨ habitat (NLCD veg class) ¨ siteID ¨ domainID All other plots:clipID ¨ plotID ¨ habitat ¨ siteID ¨ domainID The basic spaal data included in the data download include the latude, longitude, and elevaon of the centroid of the plot where sampling occurred + associated uncertainty due to GPS error and plot width. Shapefiles of all NEON TOS sampling locaons can be found in the Document Library: http://data. neonscience.org/documents. To derive a more precise esmate of the locaon of each trap, there are two opons: 1. Use the def.calc.geo.os funcon from the geoNEON package, available here: https://github.com/ NEONScience/NEON-geolocation 2. Or follow these steps to perform the same calculaon: a. The namedLocaon field in the data is the named locaon of the plot; more precise geographic data require the named locaon of the subplot. Construct the named locaon of the subplot of each record in hbp_perbout by concatenang the fields for namedLocaon and subplotID as: namedLocaon + ‘.’ + subplotID, e.g. subplotID ‘41’ of namedLocaon ‘HARV_052.basePlot.hbp’ has a complete named locaon of ‘HARV_052.basePlot.hbp.41’. b. Use the API (http://data.neonscience.org/data-api; e.g. http://data.neonscience.org/api/ v0/locations/HARV_052.basePlot.hbp.41) to query for elevaon(“locaonElevaon”), east- ing(“locaonUtmEasng”), northing(“locaonUtmNorthing”), coordinateUncertainty (“Value for Coordinate uncertainty”), elevaonUncertainty (“Value for Elevaon uncertainty”), and utmZone (“locaonUtmZone”) as inputs to the next step. c. Calculate the clipCellNumber of each trap (last 3 digits of clipID). d. Use the clip cell lookup table, available here: http://data.neonscience.org/api/v0/documents/ clipCellNumber_lookup (clicking on link will iniate download), to find the offsets for each clipCell- Number and subplot (=pointID), and use the offsets to adjust the UTM values downloaded in step 2. e. Increase coordinateUncertainty by an appropriate amount to account for error introduced by navigang within plots. Technicians may shi elevated trap up to 1 meter to either the north or south within a selected clip cell to avoid obstacles. Addional error may be introduced due to tape stretching to navigate to locaons within plots.

Page 9 of 13 Title: NEON User Guide to Herbaceous clip harvest (DP1.10023.001) Date: 05/11/2021 Author: Katherine Jones Revision: F

3.6 Temporal Resoluon and Extent

The finest temporal resoluon at which instances of the herbaceous clip harvest data product are tracked is the date of harvest, collectDate, for a given plot or subplot. collectDate (date an individual strip was clipped) ¨ eventID ¨ annual producon The NEON Data Portal provides data in monthly files for query and download efficiency. Queries including any part of a month will return data from the enre month. Code to stack files across months is available here: https://github.com/NEONScience/NEON-utilities.

3.7 Associated Data Streams

Addional informaon about woody vegetaon present in the plot, if relevant, is available in the NEON Woody Plant Vegetaon Structure data product (DP1.10098.001). Users are advised to leverage data from vegetaon structure to provide context to clip harvest data. Addional informaon about lier producon in the plot is available in the NEON Lierfall and Fine Woody Debris data product (DP1.10033.001). The herbaceous clip harvest data product is used as an input to the herbaceous producon data product, the laer of which is detailed in NEON ATBD: TOS Herbaceous Biomass – Aboveground Producv- ity Calculaons (AD[12]). Separate, addional herbaceous clip harvest sampling events are carried out to generate samples analyzed for isotopes and chemistry. The results of these analyses can be found in NEON’s plant foliar stable isotopes (NEON.DOM.SITE.DP1.10053) and plant foliar chemical properes (NEON.DOM.SITE.DP1.10051) data products.

3.8 Product Instances

There are a maximum of 2 herbaceous clip harvest bouts per year at ungrazed sites, with samples collected from no more than 40 clip strips in Tower plots and no more than 20 clip strips in Distributed plots, for a total of 60 or fewer clip strips per site. Each clip yields 1-5 herb group samples. No ungrazed site should ever exceed 500 data product instances in a single calendar year. At grazed sites, Tower plots are sampled no more than 8 mes per year. There are 2 clip strips per plot/subplot resulng in up to 80 clip strips in Tower plots. Only 1 bout per year is sorted to herb group resulng a maximum of 960 samples generated in Tower plots. In years scheduled for sampling in Dis- tributed plots, 20 clip strips yielding 1-5 mass samples each are also clipped. No grazed site should ever exceed 1160 data product instances in a single calendar year. When sampling does not occur, records are generated to account for expected plots. Obstacles to sampling such as snow or management acvies are reflected in the samplingImpraccal field. Likewise, samplingImpraccal describes instances at grazed sites when a Tower plot was not subjected to grazing during the interval prior to the sampling event and when the event is not the peak biomass sampling bout. Sampling does not occur at Distributed plots in NLCD forest types and at Distributed plots in NLCD non-forest types where herbaceous cover is less than 50 percent. These instances are reflected in the remarks field and targetTaxaPresent is set to ‘No.’

Page 10 of 13 Title: NEON User Guide to Herbaceous clip harvest (DP1.10023.001) Date: 05/11/2021 Author: Katherine Jones Revision: F

3.9 Data Relaonships

A selected clipID is only ever harvested once. The unique sampleID is generated from the clipID and the collectDate. Each sampleID created in the field sampling acvity, hbp_perbout, results in 1-5 child records, subsampleIDs, in the hbp_massdata table. Duplicates and/or missing data may exist where protocol and/or data entry abberaons have occurred; users should check data carefully for anomalies before joining tables. hbp_perbout ¨ One record expected per clipID per collectDate, producing only one instance of sampleID for all me in the hbp_perbout table hbp_massdata ¨ One record expected per sampleID per herbGroup, generates a single subampleID. Du- plicate instances of a single subampleID are expected to exist where qaDryMass = ‘Y’; these are samples that are reweighed for quality assurance purposes. sampleID and the associated barcode (sampleCode) can be used to link the tables. The sampleID and sampleCode will be generated for each collecon event, and subsampleIDs and subsampleCodes will be generated fore each funconal group within a sample in the hbp_massdata table. Physical samples will be discarded following measurement. Primary key for hbp_perbout ¨ plotID, clipID, collectDate, sampleID, sampleCode Primary key for hbp_massdata ¨ sampleID, sampleCode, subsampleID, subsampleCode, qaDryMass Data downloaded from the NEON Data Portal are provided in separate data files for each site and month requested. The neonUlies R package contains funcons to merge these files across sites and months into a single file for each table described above. The neonUlies package is available from the Compre- hensive R Archive Network (CRAN; https://cran.r-project.org/web/packages/neonUtilities/index.html) and can be installed using the install.packages() funcon in R. For instrucons on using neonUlies to merge NEON data files, see the Download and Explore NEON Data tutorial on the NEON website: https://www.neonscience.org/download-explore-neon-data.

4 DATA QUALITY

4.1 Data Entry Constraint and Validaon

Many quality control measures are implemented at the point of data entry within a mobile data entry applicaon or web user interface (UI). For example, data formats are constrained and data values controlled through the provision of dropdown opons, which reduces the number of processing steps necessary to prepare the raw data for publicaon. An addional set of constraints are implemented during the process of ingest into the NEON database. The product-specific data constraint and validaon requirements built into data entry applicaons and database ingest are described in the document NEON Raw Data In- gest Workbook for TOS Herbaceous Biomass (DP0.10023.001), provided with every download of this data product. Contained within this file is a field named ‘entryValidaonRulesForm’, which describes syntac- cally the validaon rules for each field built into the data entry applicaon. Data entry constraints are

Page 11 of 13 Title: NEON User Guide to Herbaceous clip harvest (DP1.10023.001) Date: 05/11/2021 Author: Katherine Jones Revision: F

described in Nicl syntax in the validaon file provided with every data download, and the Nicl language is described in NEON’s Ingest Conversion Language (NICL) specificaons ([AD[15]). Data collected prior to 2017 were processed using a paper-based workflow that did not implement the full suite of quality control features associated with the interacve digital workflow. A schemac of the data entry applicaon design is depicted in Figure 3.

Figure 3: Schemac of the applicaons used by ﬁeld technicians to enter herbaceous biomass data

4.2 Automated Data Processing Steps

Following data entry into a mobile applicaon of web user interface, the steps used to process the data through to publicaon on the NEON Data Portal are detailed in the NEON Algorithm Theorecal Basis Document: OS Generic Transions (AD[13]).

4.3 Data Revision

All data are provisional unl a numbered version is released; the ﬁrst release of a stac version of NEON data, annotated with a globally unique idenﬁer, is planned to take place in 2021. During the provisional period, QA/QC is an acve process, as opposed to a discrete acvity performed once, and records are

Page 12 of 13 Title: NEON User Guide to Herbaceous clip harvest (DP1.10023.001) Date: 05/11/2021 Author: Katherine Jones Revision: F

updated on a rolling basis as a result of scheduled tests or feedback from data users. The Change Log secon of the data product readme, provided with every data download, contains a history of major known errors and revisions.

4.4 Quality Flagging

The dataQF field in each data record is a quality flag for known errors applying to the record. Please see below for an explanaon of dataQF codes specific to this product.

Table 1: Descripons of the dataQF codes for quality ﬂagging

fieldName value definion Data recorded using a paper-based workflow that did not imple- dataQF legacyData ment the full suite of quality control features associated with the interacve digital workflow

Addional known quality issues that are not ﬂagged: For all records published prior to 2019-08-27, the measuredBy and recordedBy ﬁelds were populated incorrectly and may or may not be unique. Records of land management acvies, disturbances, and other incidents of ecological note that may have a potenal impact are found in the Site Management and Event Reporng data product (DP1.10111.001)

Page 13 of 13